We present a content-based query system in an image database. Queries are based on color, shape content and textual description of images. The scope of this paper is to show that without efficient synergy between image analysis techniques, more particularly mathematics, and database technology, it is very difficult to find effective, feasible and efficient solutions to the problem of retrieval in visual information systems. In this paper, we focus our presentation on the usefulness of mathematics representation of color and shape, and its integration in the database. Finally, we present the usefulness of knowledge discovery in the extraction of shared features of relevant feedback.